This discussion relates to whether an existing activity event schema exists that should be used, or whether a bespoke schema should be used that can evolve with requirements. The following is not an exhaustive list, although there does not appear to be many candidates in this area, however please let us know if there is any other schemas that may be of interest.
(a) TPTP (cbe.pdf) - "Test & Performance Tool Platform" at Eclipse: http://www.eclipse.org/tptp/
(This candidate was highlighted by Rob Cernich in a previous post).
Although this project is now archived and therefore no more work will be done on it, the event schema identifies work from a number of companies on how to structure this type of information. The implementation provided by the project was EMF based, which I personally would not want to use, however the xsd schema could be used to generate a lighter weight version.
The key point about this schema is that it is generic - there are no derived element types representing the specific events (or situations) that may occur. A particular event type would be defined in an 'extension name' field, and the associated details defined in an 'extended metadata' hierarchy.
This has pros and cons - it means that new types of event could easily be supported, however it is essentially bypassing the benefits of the schema, as the information provided for a particular 'event type' may either be wrong or missing. So schema validation would not be of any use. From a serialization perspective, the information would be more verbose, as each additional field needs to provide its name, type and value.
Looking at the schema from a simplistic perspective, it basically has three elements of interest, i) component identification, to determine what is reporting the activity, ii) context, to identify any correlation info required to relate it to other events, and iii) situation (i.e. the activity), which has a type and set of supporting name/value pair type data.
(b) CIM "Common Information Model" - (referenced by the TPTP project as well as Heiko's original SAMM project wiki page): http://dmtf.org/standards/cim
This is more management focused, so encompasses describing a managed environment, applications, but also events and metrics. So more related to interoperability of information between management systems.
My current preference is to simply define our own schema. I don't believe there is any significant benefit of strictly adopting either of the 'standard' schemas identified above, although it would be wise to follow some useful patterns followed by these schemas.
cbe.pdf 167.8 K